1
Management of Semantics in
Structured Laboratory Report
Session 132, February 13, 2019
Arianna Cocchiglia, R&D Manager, Arsenàl.IT
Myriam Ferrari, Project Engineer, Arsenàl.IT
2
Arianna Cocchiglia, R&D Manager
Myriam Ferrari, Project Engineer
Have no real or apparent conflicts of interest to report.
Conflict of Interest
3
➢ Background
➢ HIE
➢ Interoperability and focus on Lab Report
➢ Semantics Interoperability
➢ Management of Semantics
➢ Barriers and suggestions
➢ Numbers
➢ Applications and... new challenges
Agenda
4
• Employ standards in data sharing and
interoperability
• Design a working process to manage semantics
mapping activity
• Describe the functionalities of a code system to
manage semantics
Learning Objectives
5
Arsenàl.IT - Who we are
Background
6
The Healthcare System - Veneto Region
Background
“I want to tell you a story…”
One regional public health system
Nine local health authorities (LHAs)
Two hospital trusts (HTs)
One oncology research center
and...
Twenty-three different Laboratories
7
Veneto HIE System
Total number:
➢ ePrescriptions: 121,586,123
➢ eReferrals: 64,091,607
➢ Radiology reports: 2,417,812
➢ ER reports: 1,720,404
➢ Discharge letters: 559,084
➢ Outpatient visit reports: 493,224
➢ Operative reports: 276,167
➢ Lab reports: 6,507,119
8
Veneto HIE System
Why?
Secondary
usage of data
...with readable and
machine-processable
data!
Lab report:
- High number of reports available
- High number of data contained in each report
- Relevant in control and prevention (AMR
surveillance, ACG, …)
CARE
GOVERNANCE
RESEARCH
HIE
9
...and sharing means Interoperability
HIE system means, at first, sharing of eRecords...
Veneto HIE System
HIE
10
Interoperability
Starting from
HIE
11
Level 1
standard to share documents
Interoperability
HIE
12
Level 2
standard to structure documents
Interoperability
Semantics
HIE
13
Level 3
standard in language
HIE
Interoperability
Semantics
14
Level 1:
IHE Cross-Enterprise document
sharing profile (XSD.b)
Level 2:
HL7 CDA
®
R2 Structured Clinical
Document
Level 3: standard codes...
Interoperability - Lab Report
How
HIE
HIE
15
• Local terminology are
dissimilar
• Use of idiosyncratic local
codes
...the use of the same
language to define the same
concept
Interoperability - Lab Report
Semantic Interoperability
Emoglobina
glicata
Hemoglobina
glicosilada
Glycated
hemoglobin
HbA1c
Γλυκοζυλιωμέν
η αιμοσφαιρίνη
16
➢ Public
➢ Globally adopted
➢ “...LOINC fits perfectly as the vocabulary
standard for observation identifiers in
HL7 messages. In 2011, Regenstrief and
HL7 signed a statement of
understanding to enhance their long-
standing relationship to develop
complementary health data standards.”
Regenstrief/HL7 agreement (2011-11-29)
Interoperability - Lab Report
Semantic Interoperability
Standard code system LOINC
®
17
Local codes
Semantic Interoperability
The mapping process
LOINC
®
codes
1,150 laboratory tests
15,168 local test results
4,046 LOINC
®
codes
18
Semantic Interoperability
The mapping process
Local codes
LOINC
®
codes
HIE
19
➢ LHAs information system representatives
➢ Clinicians and
laboratory technicians
➢ Skillful support team
Semantic Interoperability
The multidisciplinary team
20
Semantic Interoperability
The tool
Terminology Service
...to easily perform the transcoding
from local codes to LOINC
®
codes
Transcode
21
• “co-working” between different Laboratories (suggestions from other
Labs regarding the same concept, sharing of work)
...to reduce erroneous mapping
...to accelerate mapping activity
Management of semantics
Terminology Service - main features
Show other Labs
22
• overlook of historical changes
...to better manage the work
Management of semantics
Terminology Service - main features
Restore
23
• most frequently performed lab tests are highlighted
...to have an order to perform transcoding
Management of semantics
Terminology Service - main features
24
Semantic Interoperability
Getting started
Training about:
● Terminology Service
● LOINC
®
...to increase clinicians expertise
...to reduce the undesirable mapping results
Scheduling of activities:
● cluster lab tests
● starting from the most frequently performed
… to distribute the workload!
25
➢ continuous guidance, education and counseling
➢ continuous engagement of clinicians
...to increase participation and motivation
➢ cooperation and motivation
...to overcome the “barrier to change”
Semantic Interoperability
On the job actions
26
Semantic Interoperability
On the job actions
➢ suggestions
from a support team of expert
clinicians
...to facilitate the transcoding
➢ info about progress of work
...to improve the activity growth
Progress of activities
27
➢ quality check
finding of potential erroneous mapping and suggestions
of possible correct codes
...to improve the quality of transcoding
...to enhance data quality
Semantic Interoperability
On the job actions
28
...to have
semantic enrichment
of data!
Semantic Interoperability
On the job actions
29
Laboratories periodical upload the mapped LOINC
®
codes in the LISs
Adoption of standard codes in HL7 CDA
®
R2
Laboratory Report
structured Laboratory Report are registered in
regional HIE system
Management of semantics
The final countdown
3
2
1
GO!
data sharing and
data aggregation from many sources
30
The numbers
23 Laboratories
> 50 professionals involved
∼ 20 hours per month
of activities per person
2.5 years project period
∼ 5 FTE !
31
The results
1,150 laboratory tests
15,168 local test results
4,046 LOINC
®
codes
2016
2017
2018
22 %
6 %
16 % of IMPROPER TRANSCODING
32
• local test codes ambiguous
...difficulties in quality check activity
...but laboratories had the chance to revise
configuration of their LISs system
• different level of granularity between local and
LOINC
®
codes
...introduction of other standard code systems
• “terminology service” manages only the mapping of
LOINC
®
Lab codes
...for our purpose it’s ok, but!
Barriers and taken solutions
during the project
33
• ...the positive experience could be reused in other healthcare
domains.
• improving of “Terminology Service” using CTS2
TM
Standard
...to grow the level of interoperability in semantics management
• improving of “Terminology Service” with semi-automatic methods and
machine learning algorithms
...to ease mapping and quality check activity
...to support transcoding versioning and maintenance
• “Terminology Service” integration with LIS
… to have a non-stop updating of codes in produced documents
...and
some suggestions for the future
34
• fundamental initial training phase and a constant
engagement of the clinicians
...fully gain the quality and efficiency
...to form a cohesive working team that can be
involved in other projects
• have a skillful team
...schedules activities and
performs quality check analysis of transcoding
• knowledge of shared clinical information power
...to growing the semantic enrichment and the high
quality of data
Recommendations
35
From codes to “big data”
Veneto Region
➢ interoperable system
➢ readable and processable data
➢ volume, velocity, variety, veracity, value
can be used as…
28,000 Laboratory Report per day
6,500,000 Laboratory Report accessible to be processed
...to improve medical
research and
clinical governance!
36
➢ Microbiology Lab Report:
biosurveillance system
antimicrobial resistance studies (AMR)
...go to see Session #311 on February, 15 at 12AM !
➢ Stratification of Clinical Report
based on specific geographic areas:
Adjusted Clinical Groups (ACG)
prevention programs
➢ Machine-learning algorithm
based on lab report results:
Prediction of renal function
➢ ...what else?
Applications
and new challenges
